Market research terms demystified

Jargon can make it difficult to navigate the world of market research. Here is a straight-forward guide to help you.


Analysis of variance. A test for the differences among the means of two or more variables.

Access panels

A database of individuals who have agreed to be available for surveys of varying types and topics. Rising rates of refusals and non response, make it more difficult to recruit for a single survey, therefore sampling from a pool of potentially willing marketing research respondents can be seen as an appropriate way of saving time and money.

Accompanied shopping

A specialised type of individual depth interview, which involves respondents being interviewed while they shop in a retail store and combines observation with detailed questioning.

Alternative hypothesis

The hypothesis where some difference or effect is expected (i.e. a difference that cannot occur simply by chance).

Ambiguous question

A badly constructed question which results in respondents and researchers reading different meanings into what is being asked, resulting in inappropriate or unexpected answers.


A type of stimulus material where key frames for a television advertisement are drawn or computer generated with an accompanying sound track.

Annotation method

An approach taken to analyse qualitative data using codes or comments on the transcripts to categorise the points being made by respondents.

Area sampling

A type of cluster sampling in which the clusters are created on the basis of the geographic location of the population of interest.

Audience's thinking sequence

The sequence of thoughts that people go through when they are being communicated with.


An examination and verification of the movement and sale of a product. There are three main types: wholesale audits, which measure product sales from wholesalers to retailers and caterers, retail audits, which measure sales to the final consumer, and home audits, which measure purchases by the final consumer.

Bar chart

A chart which uses a series of bars that may be positioned horizontally or vertically to represent the values of a variety of items.

Beauty parades

The procedure of asking a number of agencies to present their proposals verbally to the client company. The procedure is used to assist clients in selecting the research agency that will undertake a research project.


An abbreviated title for the term web log, meaning a frequent, chronological publication of personal thoughts and ideas. Twitter is a form of microblogging service that allows an individual to publish their blog type opinions and ideas in short Tweets
( text-based messages of up to 140 characters).

Brand Equity Modelling

Brand Equity is ultimately the strength of the brand. But there are many different views on what Brand Equity is and how to measure it within surveys. Brand Equity modelling is the creation of a Brand Equity measure, and also any Key Driver analysis to determine what drives Brand Equity (e.g. which brand imagery statements are most related to Brand Equity).

Brand Fit

Brand fit measures the fit of a brand’s profile of product qualities with the desirability of these qualities within a segment. For example, if a brand performs well on criteria that are important to a particular segment, then that brand will have a good fit with the segment. A very helpful analysis for optimising brand positioning.

Brand mapping

A projective technique which involves presenting a set of competing brand names to respondents and getting them to group them into categories based on certain dimensions such as innovativeness, value for money, service quality and product range.

Brand personalities

A projective technique which involves respondents imagining a brand as a person and describing their looks, their clothes, their lifestyles, employment, etc.

Brand Price Trade Off

Brand Price Trade Off is a technique used for establishing brand and price preferences. Respondents are presented with a set of branded products priced at the lowest possible price point for each of the brands in question. The respondent is asked to select which products they’d choose. The price of the product selected is increased to the next level, and they are then asked again which product they’d choose. This process is repeated until a product reaches its maximum price and is still selected. Competitive demand curves can be created as well as specific price scenario testing.


Computer-assisted personal interviewing. Where lap-top computers or pen-pad computers are used rather than paper-based questionnaires for face-to-face interviewing.


Computer-assisted telephone interviewing. CATI involves telephone interviewers typing respondent's answers directly into a computer-based questionnaire rather than writing them on a paper-based questionnaire.


Completely automated telephone interviews which use interactive voice technology and require no human interviewer. Respondents answer the closed-ended questions with their touch tone telephone.

CHAID (CHi-Squared Automatic Interaction Detection)

CHAID is a type of decision tree technique, based upon significance testing. It can be used to create rules to classify future respondents into identified groups using a number of different questions, or detecting interrelationships between different questions. For example, the combination of age and gender helps to further explain the different types of snacks that people consume.

Canonical Correlations

A statistical technique used to identify how much one set of independent variables (e.g. Age, Gender, Social Class) drives another set of dependent variables (e.g. Snack Choice). It is particularly useful when there are multiple dependent variables or the variables are categorical (e.g. Age, Gender etc.). It can also be used within segmentation, for example to segment on the relationship between attitudes and behaviours.

Cartoon completion

A projective technique which involves a cartoon that the respondent has to complete. For example, the cartoon may show two characters with balloons for dialogue. One of the balloons sets out what one of the characters is thinking or saying, while the other is left empty for the respondent to complete.

Causal research

Research that examines whether one variable causes or determines the value of another variable.


Research which involves collecting data from every member of the population of interest.

Chat rooms

An Internet-based facility that can be used for online focus groups where individuals are recruited who are willing to discuss a subject online usually using text.


A statistical test which tests the 'goodness of fit' between the observed distribution and the expected distribution of a variable.

Choice-Based Conjoint

Choice-Based Conjoint is a specific type of conjoint analysis where respondents are asked to make a choice between different sets of products/services, to derive the overall appeal of each component part. This is given either as a discrete choice, or as a chip-allocation style response (e.g. number out of 10 next purchases allocated to each product/service).

Closed question

A question that requires the respondent to make a selection from a predefined list of responses. There are two main types of closed questions: dichotomous questions with only two potential responses and multiple response questions with more than two.

Cluster analysis

Cluster analysis is the statistical term for the creation of segments – the process of dividing markets into groups that are similar to each other, but different to the other groups. There are a number of different ways to segment, the two most common being:

Consumer Segmentations – used to understand which consumers to target and service with distinct marketing propositions, or to tailor brands, products, pricing, communication to specific groups and make more effective use of marketing resource.

Occasion-Based Segmentations – used to understand Needs on different occasions (e.g. Had a coffee to wake up in the morning, or had a social coffee with friends after work) in order to help with new product development in repertoire markets, or for brand positioning for clients with multiple brands.

Cluster sampling

A probability sampling approach in which clusters of population units are selected at random and then all (one-stage cluster sampling) or some (two-stage cluster sampling) of the units in the chosen clusters are studied.


The procedures involved in translating responses into a form that is ready for analysis. Normally involves the assigning of numerical codes to responses.

Coefficient alpha

See Cronbach alpha

Coefficient of determination

Measure of the strength of linear relationship between a dependent variable and independent variables.

Concept boards

A type of stimulus material which uses a set of boards to illustrate different product, advertising or pack designs.

Confidence level

The probability that the true population value will be within a particular range (result +/– sampling error).

Conjoint analysis

A statistical technique that provides a quantitative measure of the relative importance of one attribute over another. It is frequently used to determine what features a new product or service should have and also how products should be priced.

Conjoint analysis

An abbreviation of “consider jointly”, Conjoint Analysis is a powerful statistical technique to understand what combination of a limited number of attributes or features is most influential in the consumer’s decision-making process.

Conjoint Analysis is a multivariate technique based on the idea that when choosing a product or service, people will trade off features simultaneously. This process is replicated within a respondent exercise – a number of scenarios are shown where respondents are asked to express choice or preference amongst a set of products/services comprising of different combinations of attributes levels. A model is created from the resulting data to simulate “what if” scenarios and to assess to impact of each attribute in the decision-making process. Conjoint analysis is often used in concept testing studies, and pricing research.

Construct validity

An analysis of the underlying theories and past research that supports the inclusion of the various items in the scale. It is most commonly considered in two forms, convergent validity and discriminant validity.

Content analysis

The analysis of any form of communication, whether it is advertisements, newspaper articles, television programmes or taped conversations. Frequently used for the analysis of qualitative research data.

Content analysis software

Software used for qualitative research which basically counts the number of times that pre-specified words or phrases appear in text.

Content validity

A subjective yet systematic assessment as to how well a rating scale measures a topic of interest. For example a group of subject experts may be asked to comment on the extent to which all of the key dimensions of a topic have been included.

Continuous research

See Longitudinal research.

Contrived observation

A research approach which involves observing participants in a controlled setting.

Convenience sampling

A non-probability sampling procedure in which a researcher's convenience forms the basis for selecting the potential respondents (i.e. the researcher approaches the most accessible members of the population of interest).

Convergent validity

A measure of the extent to which the results from a rating scale correlate with those from other scales or measures of the same topic/construct.


Text files placed on a user's computer by web retailers in order to identify the user when he or she next visits the website.


A statistical approach to examine the relationship between two variables. Uses an index to describe the strength of a relationship.

Critical path method (CPM)

A managerial tool used for scheduling a research project. It is a network approach that involves dividing the research project into its various components and estimating the time required to complete each component activity.

Cronbach Alpha

A statistical test used to measure the split half reliability of a summated rating scale. Also known as coefficient alpha.

Customer database

A manual or computerised source of data relevant to marketing decision making about an organisation's customers.

Cut and paste method of analysis

A method for analysing qualitative research data where material is cut and pasted from the original transcript into separate sections or tables relating to each topic. Cutting and pasting can either be done physically using scissors or using a word processing computer package.

Data analysis errors

Non-sampling errors that occur when data is transferred from questionnaires to computers by incorrect keying of information.

Discriminant analysis

Discriminant analysis uses the responses to a set of questions (e.g. attitudes) to predict existing group membership (e.g. segments). The output can then be used to classify future respondents into the same groups using their responses to the same set of questions. Similar to CHAID, but used when all independent variables are scales.


The process of ensuring that questionnaires were filled out correctly and completely

Face-to-face survey

Research which involves meeting respondents face-to-face and interviewing them using a paper-based questionnaire, a lap-top computer or an electronic notepad.

Factor Analysis

Factor Analysis is a statistical technique to examine the similarities between items in order to identify a more concise summary of themes. For example, from a list of 20 statements on car imagery, we may identify a factor on reliability, design, performance, environment and image. This can be useful for ordering statements in a presentation so that similar statements are presented together, or for data reduction purposes e.g. to focus a segmentation or to take account of bias in questionnaire design that included too many statements from one particular theme.

Gabor Granger

Gabor Granger is a pricing technique used to understand price elasticity for set products. Respondents are asked how likely they are to purchase a product at a number of different price points. The purchase intention measures can be converted to estimated take-up scores and plotted to establish which price point is most suitable. The technique can be extended to present products within a competitive context.

GANTT chart

A managerial tool used for scheduling a research project. It is a form of flowchart that provides a schematic representation incorporating the activity, time, and personnel requirements for a given research project.

Hall tests

Research undertaken in a central hall or venue commonly used to test respondents' initial reactions to a product or package or concept. Respondents are recruited into the hall by interviewers stationed on main pedestrian thoroughfares nearby.

Impact indices

Impact indices measure the impact any independent variable has on changing a dependent variable. Often used when the independent variables are binary (e.g. Yes/No), it can be used for example to evaluate the impact of different product qualities on preference for the product.

Implicit assumption

A badly constructed question where the researcher and the respondent are using different frames of reference as a result of assumptions that both parties make about the question being asked.

Judgement sampling

A non-probability sampling procedure where a researcher consciously selects a sample that he or she considers to be most appropriate for the research study.

Key Driver Analysis

The analysis of the relationship between a dependent variable (e.g. brand strength) and one or more independent variables (e.g. brand imagery statements). Its purpose is to determine whether a relationship exists and the strength of the relationship, and used to help prioritise what to focus on.

There are many different statistical techniques that fall under this term, from correlations to Structural Equation Modelling. Different techniques can be applied depending on the objectives, the type of data and how the results will be used.

Kiosk based Survey

A survey often undertaken at an exhibition or trade show using touch screen computers to collect information from respondents. Such computers can be programmed to deliver complex surveys supported by full colour visuals as well as sound and video clips. They can be much cheaper to administer in comparison with the traditional exit survey undertaken by human interviewers.

Kruskal's relative importance analysis

A type of Key Driver Analysis, Kruskals’ relative importance analysis is as an alternative to other techniques such as ordinary regression analysis, which can give misleading results when there is missing data, or when variables are strongly related to each other (which is typical of research data).

Leading question

A badly constructed question that tends to steer respondents toward a particular answer. Sometimes known as a loaded question.

Linear regression

Linear regression is used to find out the relative importance of different drivers in order to re-create a dependent variable. For example, the influence of brand imagery items onto brand appeal.

Logistic regression

Logistic regression is used to find out the relative importance of different drivers in order to re-create a dependent variable when the dependent variable is binary (e.g. Yes/no or Buy/not buy). It is used when the usual linear regression cannot be used, and is particularly useful in propensity modelling.

MBC (Menu-Based Conjoint)

Menu-based conjoint is a specific type of conjoint analysis able to handle a variety of menu choice situations in which respondents make from one to multiple choices in the process of building their preferred selection. An example situation to which this would apply would be a fast-food restaurant where it’s possible to choose something from the ‘fixed-menu’ section, with some personalisation e.g. choice of side dish - or instead purchasing a series of single items.

Mall intercept interviews

See Street interviews.

MaxDiff (Maximum Difference Scaling)

MaxDiff is a technique used to understand relative importance or appeal amongst a list of features/statements. Respondents are asked to compare sets of typically 4-5 features or statements stating which of these are the most and least appealing/important to them. The results give a % appeal/importance score for each item, and can be used to identify which are most popular. Often used to force differentiation, when simple scales may not work as well (e.g. consumers thinking all attributes are important).


Sometimes known as online ethnography and webnography is the ethnographic study of communities on the World Wide Web. It generally involves a researcher fully participating as a member of the online community.


A data gathering approach where information is collected on the behaviour of people, objects and organisations without any questions being asked of the participants.

Open-ended question

Questions which allow respondents to reply in their own words. There are no pre-set choices of answers and the respondent can decide whether to provide a brief one-word answer or something very detailed and long. Sometimes known as unstructured questions.

Paired interviews

An in-depth interview involving two respondents such as married couples, business partners, teenage friends or a mother and child.

Price Sensitivity Management (Van Westendorp)

Price Sensitivity Measurement (PSM) is a technique used to understand price preferences. Respondents are asked at what price they would consider a product to be:

• too expensive for them to consider it
• too cheap so they would start to doubt its quality
• expensive but would still consider buying it
• a bargain or good value for money

The data from these questions is plotted across the sample to see where the cumulative frequencies intersect. For example the optimal price point is determined to be where an equal proportion of respondents have said “too cheap” and “too expensive”.

Qualitative research

An unstructured research approach with a small number of carefully selected individuals used to produce non-quantifiable insights into behaviour, motivations and attitudes.


A measure of dispersion that calculates the difference between the largest and smallest values in a set of data


A subset of the population of interest.


The process of dividing markets into groups with people or occasions that are similar to each other, but different to the other groups.

There are a number of different ways to segment. The two most common are:

Consumer Segmentations – used to understand which consumers to target and service with distinct marketing propositions, or to tailor brands, products, pricing, communication to specific groups and make more effective use of marketing resource.

Occasion-Based Segmentations – used to understand Needs on different occasions (e.g. Had a coffee to wake up in the morning, or had a social coffee with friends after work) in order to help with new product development in repertoire markets, or for brand positioning for clients with multiple brands.

Structural Equation Modelling

Structural Equation Modelling (SEM) is a statistical technique for testing and estimating causal relationships, using a combination of statistical data and qualitative causal assumptions. Factor analysis, path analysis and regression all represent techniques used within SEM, which also allows the construction of variables which are not measured directly. As an example, it can be used to model and understand the relationship between different aspects of customer satisfaction and how these explain customer loyalty.

t test

A hypothesis test about a single mean if the sample is too small to use the Z test.


TURF stands for Total Unduplicated Reach and Frequency, and is used for providing estimates of media or market potential and devising optimal communication and placement strategies. If we take appeal of ice cream flavours as an example, TURF analysis can identify the number of users reached by the combination of ice cream flavours in a client’s range, as well as how frequently they will be consumed - very useful for deciding which flavours of ice cream to have in a range.


See Population of interest.

Unstructured questions

See Open-ended questions.

User generated content

Online material such as comments, profiles photographs that is produced by end users.


Whether the subject requiring to be measured was actually measured.

Video conferencing

The bringing together of a group of individuals using a video link and telecommunications. Can potentially be used for group discussions, particularly where the respondents are located in various parts of the world.

Viewing rooms

Specialist facilities/locations for group discussions. They are set out in the form of a boardroom or living-room setting with video cameras or a large one-way mirror built into one wall. Some are owned by research agencies, but the majority are independent and available to anyone willing to pay the hourly room-hire rates.

Web 2.0

Web applications that facilitate interactive information sharing, interoperability and collaboration on the Internet.

Web survey

see Online survey


The process of adjusting the value of survey responses to account for over- or under-representation of different categories of respondent. Weighting is used where the sample design is disproportional or where the achieved sample does not accurately reflect the population under investigation.

Word association tests

A projective technique that involves asking respondents what brands or products they associate with specific words. In addition to the direct outputs regarding brand imagery, it is also a very useful technique for building rapport within a group discussion and getting everybody contributing and involved.

Word cloud

A visual depiction of words used by respondents in qualitative research or the content appearing on social network sites or publications. The font size of the words, is determined by the number of times a word has been used. The more that a topic or word is talked about, the bigger it is.

Z test

A hypothesis test about a single mean where the sample size is larger than 30.

Brought to you by